Security In The Software Development Lifecycle · Security In The Software Development Lifecycle Hala Assal & Sonia Chiasson School of Computer Science, Carleton University Ottawa

Security In The Software Development Lifecycle

Hala Assal & Sonia ChiassonSchool of Computer Science, Carleton University

Ottawa ON, [email protected]

1

mailto:[email protected]

Why?

• Vulnerabilities persist

• Increasing and evolving threats

• Developers are not the enemy

https://www.forbes.com

2

Related Work• Security tools’ adoption

• E.g., Witschey et al., 2014; Xiao et al., 2014; Xie et al., 2011

• Proposing and improving new tools and methodologies

• E.g., Smith et al., 2018; Nguyen et al., 2017; Xie et al., 2011

• Developers’ abilities and experience

• E.g., Smith et al., 2015; Oliveira et al., 2014; Baca et al., 2009 3

The overall security processExisting practices compared to best practices

4

Approach

5

Interview studyExplore existing

practicesAmalgamate best

practices

Compare practicesExplore influential

factors

Study Design

• Semi-structured interview

• Software engineering + security topics

• ~1hr sessions

• Audio recorded and transcribed

6

Participants

• 13 professional developers

• Avg. experience = 9.35 years (Md=8)

• 15 teams in 15 different companies

• 7 SMEs and 8 large enterprises

7

Analysis Methodology

• Qualitative Content Analysis (Elo & Kyngäs, 2008)

• Deductive and inductive approaches

• Analysis Matrix of development activities

• 264 unique excerpts

8

Selectingtheunitofanalysis

Makingsenseofthedataandobtainasenseofwhole

Opencoding

Codingsheets

Grouping

Categorization

Abstraction

Developingstructuredanalysis

matrix

Datacodingaccordingtothecategories

Hypothesistestingandcomparisontoearlierstudies

Developinganalysismatrix

Datagatheringbycontent

Model,conceptualsystemconceptualmaporcategories

Deductive approach Inductive approach






9



Opencoding

Codingsheets

Grouping

Categorization

Abstraction


matrix












10



Opencoding

Codingsheets

Grouping

Categorization

Abstraction


matrix












11



Opencoding

Codingsheets

Grouping

Categorization

Abstraction


matrix












12



Opencoding

Codingsheets

Grouping

Categorization

Abstraction


matrix







Security In Practice

13

The Security Adopters & The Security Inattentive

SecurityAdopters SecurityInattentive

:secure, :somewhatsecure, :notsecure, :notperformed, :nodata 14

Table 2: Summary of themes emerging from the security adopters and the security inattentive, and common themes betweenthe two groups. Although common themes exist, driving factors for these themes may di↵er. See Section 4.2 for more details.

Security Adopters Themes Common Themes Security Inattentive Themes

Design

· Security design is very important · Security is not considered in the designstage

· Security consideration in the design stage is adhoc

Implementation

· Security is a priority during implementa-tion

· Developers’ awareness of security is ex-pected when implementing

· Security is not a priority during implementation· Developers take security for granted· Developers misuse frameworks· Developers lack security knowledge· Developers perceive their security knowledge inaccurately· Vulnerability discovery can motivate security

Developer Testing

· Developers test for security fortuitously· Security is a priority during developer testing

· Developers do not test for security · Developers’ security testing is feature-driven

Code Analysis

· Security is a priority during code analysis · Security is a secondary objective during code analysis· Developers rarely perform code analysis, never for security· Developers vary in awareness of analysis tools

Code Review

· Code review is a formal process that includes security· Preliminary code review is done as a checkpoint be-fore the formal review

· Security is not considered during code review· Security consideration in code review is minimal

Post-development Testing

· Security is a priority during post-development testing· Post-development testing is used to discover securityvulnerabilities, or for final verification

· Testers have the final approval

· Security is not considered in post-development testing· Post-development testing plans include a security dimension· Post-development security testing is feature-driven· Post-development security testing is adhoc· Post-development security testing is externally-driven

and developers, whereas the Testing Guide [4] focuses onbest practices for testing and evaluating security activities.

Others. Additional resources for security best practices in-clude: NASA’s Software Assurance Guidebook [39], NIST’sSpecial Publication 800-64 [32], US-CERT’s Top 10 SecureCoding Practices [47], as well as various articles emphasizingthe importance of secure development [7, 36,37,57].

5.2 Security Best PracticesAvailable resources for security best practices vary in theirorganization and their presentation style, e.g., they vary intechnical details. Practitioners may find di�culty decidingon best practices to follow and establishing processes withintheir organizations [38,42,54]. To help frame security prac-tices we identified, we collected recommendations from thesources discussed in Section 5.1 to compose a concise set ofbest practices. This resulted in an initial set of 57 unor-ganized recommendations varying in format and technicaldetails. We then grouped related recommendations, orga-nized them in high-level themes, and iterated this process tofinally produce the following 12 best practices. Other amal-gamations may be possible, but we found this list helpful tointerpret our study results. The list could be of independentinterest to complementary research in this area.

B1 Identify security requirements. Identify security re-quirements for your application during the initial planningstages. The security of the application throughout its dif-ferent stages should be evaluated based on its compliancewith security requirements.B2 Design for security. Aim for simple designs because

the likelihood of implementation errors increases with de-sign complexity. Architect and design your software to im-plement security policies and comply with security princi-ples such as: secure defaults, default deny, fail safe, andthe principle of least privilege.B3 Perform threat modelling. Use threat modelling toanalyze potential threats to your application. The resultof threat modelling should inform security practices in thedi↵erent SDLC stages, e.g., for creating test plans.B4 Perform secure implementation. Adopt secure cod-ing standards for the programming language you use, e.g.,validate input and sanitize data sent to other systems, andavoid using unsafe or deprecated functions.B5 Use approved tools and analyze third-party tools’

security. Only use approved tools, APIs, and frameworksor those evaluated for security and e↵ectiveness.B6 Include security in testing. Integrate security testingin functional test plans to reduce redundancy.B7 Perform code analysis. Leverage automated toolssuch as SATs to detect vulnerabilities like bu↵er overflowsand improper user input validation.B8 Perform code review for security. Include securityin code reviews and look for common programming errorsthat can lead to security vulnerabilities.B9 Perform post-development testing. Identify secu-rity issues further by using a combination of methods, e.g.,dynamic analysis, penetration testing, or hiring externalsecurity reviewers to bring in a new perspective.

B10 Apply defense in depth. Build security in all stagesof the SDLC, so that if a vulnerability is missed in onestage, there is a chance to eliminate it through practicesimplemented in the remaining stages.

15



Design



Implementation




Developer Testing



Code Analysis


Code Review














16

Security Adopters

Security Inattentive



Design



Implementation




Developer Testing



Code Analysis


Code Review














17

Common



Design



Implementation




Developer Testing



Code Analysis


Code Review














18


Design



Implementation




Developer Testing



Code Analysis


Code Review







19

Security is a priority during post-development Security is not considered in post-development



Design



Implementation




Developer Testing



Code Analysis


Code Review














20

–Adopters theme: Security design is very important

“When we go to do a further security analysis, we have a lot more context in terms of what we’re

thinking, and people aren’t running around sort of defending threats that aren’t there. ”

21

– Inattentive theme: Developer testing is feature-driven

“Security testing [pause] I would say less than 5%. Because we’re doing embedded systems, so security

[is] pretty low in this kind of work.”

22

Digging Deeper…

23

• Division of labour• Resource availability• Security knowledge

• Company culture• External pressure• Experiencing a security incident

Factors Affecting Security Practices

24

Factors Affecting Security Practices• Division of labour• Resource availability• Security knowledge


25

• Division of labour• Resource availability• Security knowledge


Factors Affecting Security Practices

26

“Probably in the two years that I’ve been working, I never got feedback [on] the security of my code [...]

[Developers reviewing code] don’t pay attention to the security aspect and

they can’t basically make a comment about the security of your code.”



27



28

“No one had really been thinking about looking at the product from security

stand-point and so the new tester we had hired, he really went at it from ‘how can I really break this thing?’ [..] and found quite

a few problems with the product that way.”

Now What?

29

Now What?

• Rethink best practices

• Focus on building a security company culture

30

“To be honest, I don’t think anybody cares about [security]. I’ve never heard or seen people talk

about security at work [...] I did ask about this to my managers, but they just said ‘well, that’s how the

company is. Security is not something we focus on right now.’”

–Factor: Company culture

31

Documents

Security In The Software Development Lifecycle · Security In The Software Development Lifecycle Hala Assal & Sonia Chiasson School of Computer Science, Carleton University Ottawa